Soft Tactile Fingertip to Estimate Orientation and the Contact State of Thin Rectangular Objects
Soft Tactile Fingertip to Estimate Orientation and the Contact State of Thin Rectangular Objects
复制标题
DOI：
10.1109/lra.2019.2950118
复制
发表时间：
2020-01
影响因子：
5.2
通讯作者：
Muhammad Hisyam Rosle;R. Kojima;Keung Or;Zhongkui Wang;S. Hirai
中科院分区：
文献类型：
--
作者：
Muhammad Hisyam Rosle;R. Kojima;Keung Or;Zhongkui Wang;S. Hirai
开发了一种柔软的触觉指尖，以提供关于用于执行电子部件的自动组装的对象（诸如薄电路板）的特性的接触反馈。然而，设计具有简单设计和最少数量的传感器的触觉指尖，可以估计物体的方向和接触状态是一项具有挑战性的任务。介绍了一种结构简单、对外力敏感的柔性指尖。指尖可以估计薄矩形物体的抓取力和方向，以及基于磁通密度（MFD）变化的物体与环境之间的接触状态。两个霍尔效应传感器平行固定在硬基座上，四个圆柱形磁体嵌入指尖的软体中。除磁铁外的所有部件都是3D打印的，以简化制造过程。我们应用机器学习的方法来定义传感器输出的MFD和对象的特性之间的关系。此外，有限元模拟进行评估所设计的结构。实验结果表明，所提出的指尖可以估计薄矩形物体的方向和夹持力。此外，它可以成功地分类的接触状态的基础上的传感器输出的被抓物体。
A soft tactile fingertip was developed to provide the contact feedback on characteristics of an object for performing automatic assembly of electronic parts, such as a thin circuit board. However, designing a tactile fingertip with simple design and minimal number of sensors that can estimate orientation and the contact state of an object is a challenging task. This letter presents the simple-structured soft fingertip sensitive to external forces. The fingertip can estimate the grasping force and orientation of a thin rectangular object, and the contact state between the object and the environment based on magnetic flux density (MFD) changes. Two Hall-effect sensors were fixed in parallel at the hard base, and four cylindrical magnets were embedded in the soft body of the fingertip. All parts except the magnets were 3D-printed to simplify the fabrication process. We applied a machine learning approach to define the relationship between the MFD of sensor outputs and object's characteristics. In addition, finite element simulations were performed to evaluate the designed structure. Experimental results verified that the proposed fingertip can estimate the orientation and gripping force of thin rectangular objects. Furthermore, it can successfully classify contact states of the grasped object based on the sensor outputs.